Artificial Analysis · CritPt
The Frontier
Best score over time · one chart, every benchmark
Full rankings
47 models tested · sorted by score
| # | Model | Score |
|---|---|---|
| 1 | 31.7 | |
| 2 | 31.7 | |
| 3 | 31.7 | |
| 4 | 31.4 | |
| 5 | 30.0 | |
| 6 | 29.7 | |
| 7 | 27.1 | |
| 8 | 26.6 | |
| 9 | 23.4 | |
| 10 | 20.9 | |
| 11 | 20.0 | |
| 12 | 19.4 | |
| 13 | 19.1 | |
| 14 | 18.9 | |
| 15 | 18.3 | |
| 16 | 18.0 | |
| 17 | 17.7 | |
| 18 | 17.7 | |
| 19 | 17.7 | |
| 20 | 16.9 | |
| 21 | 15.4 | |
| 22 | 14.3 | |
| 23 | 12.0 | |
| 24 | 10.6 | |
| 25 | 10.0 | |
| 26 | 9.1 | |
| 27 | 5.4 | |
| 28 | 5.4 | |
| 29 | 4.9 | |
| 30 | 3.7 | |
| 31 | 3.1 | |
| 32 | 3.1 | |
| 33 | 2.6 | |
| 34 | 1.4 | |
| 35 | 1.4 | |
| 36 | 1.4 | |
| 37 | 1.1 | |
| 38 | 1.1 | |
| 39 | 0.9 | |
| 40 | 0.9 | |
| 41 | 0.9 | |
| 42 | 0.6 | |
| 43 | 0.3 | |
| 44 | 0.3 | |
| 45 | 0.3 | |
| 46 | 0.3 | |
| 47 | 0.3 |
Score distribution
Where models cluster
Correlated benchmarks
Pearson r · original research
Benchmarks that track with Artificial Analysis · CritPt
Pearson correlation across models scored on both benchmarks. Closer to 1 = strongly predictive.
Frequently asked
About Artificial Analysis · CritPt
What does Artificial Analysis · CritPt measure?
Artificial Analysis · CritPt is a knowledge benchmark in the BenchGecko catalog. 47 AI models have been tested on it. Scores range from 0.3 to 31.7 out of 100.
Which model leads on Artificial Analysis · CritPt?
Claude Opus 5.5 from Anthropic leads Artificial Analysis · CritPt with a score of 31.7. The median score across 47 tested models is 10.6.
Is Artificial Analysis · CritPt saturated?
No · the top score is 31.7 out of 100 (32%). There is still meaningful room for improvement on Artificial Analysis · CritPt.
Does Artificial Analysis · CritPt predict performance on other benchmarks?
Yes · Artificial Analysis · CritPt scores correlate 0.97 with Terminal Bench across 6 shared models. Models that do well on Artificial Analysis · CritPt tend to do well on Terminal Bench.
How often is Artificial Analysis · CritPt data refreshed?
BenchGecko pulls updates daily. New model scores on Artificial Analysis · CritPt appear as soon as they are published by Epoch AI or the model provider.
- Category
- Knowledge
- Max score
- 100
- Models
- 47
- Updated
- 2026-10-13
Top on Artificial Analysis · CritPt
Claude Opus 5.5 · 31.7GPT-6 Astra · 31.7GPT-6.1 Sol · 31.7Claude Sonnet 5.5 · 31.4GPT-5.6 Terra · 30.0More knowledge benchmarks
Same category · related evaluations